Re: time in lib

From: Andres Kroonmaa <andre@dont-contact.us>
Date: Fri, 4 May 2001 01:07:29 +0200

On 4 May 2001, at 0:28, Henrik Nordstrom <hno@hem.passagen.se> wrote:

> But on the other hand.. measurements have shown that gettimeofday() is a
> very quick call, at least on Linux.. think it was somewhere around 60 ns
> on my 450MHz Pentium III laptop... (well, this was ofcourse from a loop
> all cached in L1 cache but...)

 Usually it is. But this depends on hell knows what. OS/hardware combination
 could mean alot. I remember Sparcs on solaris 2.5 or so were bastardly slow.
 And now, having Solaris7 x86 on PII-450 shows 4.6us best, 5us avg, and upto
 20us worst case. Dual PIII-800 and Solaris8 shows consistently 900ns.
 I guess its to do with threading support, it could be a thread cancellation
 point, maybe passes over mutex or two, maybe passes to kernel, etc.
 trace -c on Squid hints that this call accounts for about 15-20% of total
 system time of squid. go figure.

------------------------------------
 Andres Kroonmaa <andre@online.ee>
 CTO, Delfi Online
 Tel: 6501 731, Fax: 6501 708
 Pärnu mnt. 158, Tallinn,
 11317 Estonia
Received on Thu May 03 2001 - 17:12:07 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:58 MST